Table 3 Association between human-extracted image phenotype and computer-extracted image phenotype
From: Breast MRI radiomics: comparison of computer- and human-extracted imaging phenotypes
Human-extracted assessment | Computer-extracted feature | p Value |
|---|---|---|
Lesion size | Effective diameter | <10−12 |
Surface area-to-volume ratio | <10−12 | |
Maximum linear size | <10−12 | |
Lesion volume | <10−12 | |
Shape | Sphericity | 5.65 × 10−5 |
Irregularity | 0.00398 | |
Surface area-to-volume ratio | 0.363 | |
Internal enhancement | Contrast | 0.378 |
Correlation | 0.409 | |
Difference in entropy | 0.353 | |
Difference variance | 0.191 | |
Energy | 0.186 | |
Entropy | 0.194 | |
Inverse difference moment | 0.466 | |
IMC1 | 0.328 | |
IMC2 | 0.340 | |
Maximum correlation coefficient | 0.336 | |
Sum average | 0.151 | |
Sum entropy | 0.336 | |
Sum variance | 0.702 | |
Sum of squares | 0.297 | |
Degree of margin spiculation | Mean margin sharpness | 0.292 |
Variance margin sharpness | 0.227 | |
Variance radial gradient histogram | 0.055 |